Research on Statistical Color Space Partitioning for Image Annotation

• Main Authors: Yuan-ming Yeh, 葉元明
• Other Authors: Yi-leh Wu
• Format: Others
• Language: en_US
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/91643025208743190664

碩士 === 國立臺灣科技大學 === 資訊工程系 === 97 === 　　For visual information management, image annotation which refers to the labeling of images with a set of predefined keywords is mainly used in a variety of domains such as web image classification, search, military, biomedicine, etc. 　　The Border/Interior pixel Classification (BIC) features [15] are very efficient and compact features that capture the information of color, shape, and texture. But the BIC features inherit the problem that the utilization rates are not balanced. To overcome this problem, we propose to employ the Hilbert-Scan method and the One-pass Partitioning Method (OPM). Finally, we show the accuracy by our proposed method with KNN and SVMs in annotating 6000 images in 60 categories.